Technical Field
The present invention relates to a technique for detecting an object using an image and a technique for performing vehicle control using the detected object.
Related Art
In a known technique, various objects that are present around the vehicle are detected using picked-up images that are acquired from an in-vehicle camera. In the technique, the detection results are provided to the vehicle driver or used for various vehicle control processes.
If an imaged scene has a high contrast, the image has overexposure areas or underexposure areas due to the insufficient dynamic range of the camera. These overexposure areas or underexposure areas may prevent detection of the objects that are present in the areas.
On the other hand, in another known technique, a plurality of images that are picked up under different exposure conditions are acquired, and image recognition is conducted using portions having good contrast of the plurality of images (e.g., see UP-A-2012-088785).
The conventional art as mentioned above uses a plurality of images that are acquired at different time points and therefore it is necessary to determine whether or not the objects detected in the individual images are identical. The processing involved in this determination problematically takes time. In addition to this, the conventional art suffers from problems, such as that a single object may be erroneously detected as being a plurality of different objects, or different objects may be erroneously detected as being a single object.